Laplacian Unwrapping Issue

[Matthew Hoch]

Hello All, 

Does anyone else have the issue of the Laplacian unwrapping function producing an iFreq matrix of all zeros? I am not sure how to mitigate this? Looking at the code: 

if matrix_size(3)>1

h=( (X==0) & (Y==0)&(Z==0) ) .*1;

k=6 * del2(h,voxel_size(1), voxel_size(2), voxel_size(3));

kernel=fftn(fftshift(k));

else

It looks like it sets all of the h values to 0 before calculating the kernel, is this correct? Any help would be greatly appreciated. 

Thank you, 

Matt

 

[spince...@gmail.com]

Matt,

What is your matrix_size and voxel_size?

Pascal

[Matthew Hoch]

Thank you for your help! 

matrix_size= [188, 186, 81]

voxel_size=[0.2606, 0.2606, 0.2606] (mm) 

[spince...@gmail.com]

This is indeed a bug triggered by the odd number of slices. Add a slice of zeros to your iField after you read it with Read_DICOM to make it even. Something like

iField=padarray(iField, [ 0 0 1 0], 'post');

matrix_size(3)= matrix_size(3)+1;

You can remove the extra slice after the MEDI_L1 call to compute QSM.

[Matthew Hoch]

That worked! Thank you very much!